April 5, 2018

Issues to Keep in Mind

  • How to equate a Chessie BIBI rating to stream miles
    • One sampling event equals X amount of miles?
    • Aggregate by a spatial unit (e.g., HUC 12) and use sampling events within a spatial unit cell to represent all stream miles within that spatial unit cell?
  • Spatial Influence
    • Sampling density and frequency (Sampling Rounds)
    • The amalgamation of two or more random sampling designs does NOT equate to a single random sampling design
    • Stream access, generally, leads to a larger proportion of streams sampled in urban areas
  • Temporal Influence
    • Sampling density and frequency (Sampling Rounds)
    • 2000-2008 vs. 2004-2008 baseline periods (Other Suggestions?)
  • Index Resolution
    • Region indices vs. Bioregion indices

Six Methods

  1. No Spatial Aggregation
  2. Watershed Mean Score
  3. Ratings Weighted by Stream Mile
  4. Random Sample
  5. Bootstrap
  6. Random Forest

No Spatial Aggregation

Process

  1. The proportion of Acceptable, Fair, and Degraded sampling events found in the basin

No Spatial Aggregation

No Spatial Aggregation

Discussion

  • Simplest approach (CBP Method)
  • Spatial bias
    • Frequency and density of samples
  • How to convert to stream mile units?
    • Should each point represent one stream mile?
    • Should stream order influence the number of stream miles represented by a point?
    • If the data is mapped, should the representative stream miles be drawn above the sampling point or be split by the sampling point?
      • What if two or more points have overlapping representative stream miles?

No Spatial Aggregation

No Spatial Aggregation

Watershed Mean Score

Process

  1. Chessie BIBI scores are aggregated by a spatial unit (HUC8, HUC10, HUC12, or catchment)
    • What is an appropriate number of samples to accurately represent a spatial unit cell?
    • Any spatial unit cells without the appropriate number samples are classified as "Insufficient"
  2. The mean Chessie BIBI score is calculated for each spatial unit cell
  3. Each spatial unit cell is assigned a rating based on the mean score
  4. Find the number of stream miles represented in each spatial unit cell
  5. The spatial unit cell rating is weighted by the number of stream miles
  6. Summarize the proportion of stream miles representing Acceptable, Fair, Degraded, and Insufficient at the Chesapeake Bay basin level

Watershed Mean Score

Watershed Mean Score

Watershed Mean Score

Discussion

Watershed Mean Score

Ratings Weighted by Stream Mile

Process

  1. Chessie BIBI samples are aggregated by a spatial unit (HUC8, HUC10, HUC12, or catchment)
    • What is an appropriate number of samples to accurately represent a spatial unit cell?
    • Any spatial unit cells without the appropriate number samples are classified as "Insufficient"
  2. Find the number of stream miles represented in each spatial unit cell
  3. Divide the number of stream miles in a spatial unit cell by the number of Chessie BIBI samples reported in the cell
  4. Use the divided stream miles as a weighting factor for each Chessie BIBI rating
  5. Summarize the proportion of stream miles representing Acceptable, Fair, Degraded, and Insufficient at the Chesapeake Bay basin level

Ratings Weighted by Stream Mile

Ratings Weighted by Stream Mile

Ratings Weighted by Stream Mile

Dissucsion

Ratings Weighted by Stream Mile

Ratings Weighted by Stream Mile

Ratings Weighted by Stream Mile

Random Sample

Process

  1. Chessie BIBI scores are aggregated by a spatial unit (HUC8, HUC10, HUC12, or catchment)
    • Any spatial unit cells without the appropriate number samples are classified as "Insufficient"
  2. A random sample with replacement of spatial units
  3. Find the number of stream miles represented in each spatial unit cell
  4. The spatial unit cell rating is weighted by the number of stream miles
  5. Summarize the proportion of stream miles representing Acceptable, Fair, Degraded, and Insufficient at the Chesapeake Bay basin level

Random Sample

Random Sample

Random Sample

Discussion

Random Sample

Bootstrap

Process

  1. Chessie BIBI scores are aggregated by a spatial unit (HUC8, HUC10, HUC12, or catchment)
    • Any spatial unit cells without the appropriate number samples are classified as "Insufficient"
  2. Iteratively perform a random sample with replacement of spatial units
  3. Find the number of stream miles represented in each spatial unit cell
  4. For each iteration, spatial unit cell rating is weighted by the number of stream miles
  5. For each iteration, summarize the proportion of stream miles representing Acceptable, Fair, Degraded, and Insufficient at the Chesapeake Bay basin level
  6. Find the mean and standard deviation of the proportion of stream miles representing Acceptable, Fair, Degraded, and Insufficient at the Chesapeake Bay basin level for all iterations

Bootstrap

Bootstrap

Bootstrap

Discussion

Bootstrap

Random Forest

Process

  1. Chessie BIBI ratings are aggregated by catchment
    • 1:24,000 high resolution SHEDS
  2. A random forest model was developed to predict Chessie BIBI ratings based on the geospatial predictor variables (e.g., land cover, soil, precipitation, and number of dams)
  3. Using the model predict across the entire Chesapeake Bay basin if an area represents FairGood, Poor, or Uncertain

Random Forest

Random Forest

Discussion

Random Forest

—- {.flexbox .vcenter}

Break for Lunch

12:00-1:00